Kernel-based tracking using spatial structure

We extend the concept of kernel-based tracking by modeling the spatial structure of multiple tracked feature points belonging to the same object by a simple graph-based representation. The task of track-ing parts or multiple feature points of an object without considering the underlying structure becomes ambiguous if the target representation (for example color histograms) is similar to other nearby tar-gets or to that of the background. Instead of considering tracking of multiple targets as isolated processes, we propose an approach incorporating spatial dependencies between tracked targets and an iterative technique to efficiently locate the spatial arrangement of targets maximizing the joint pos-terior. We present a series of experiments demonstrating that the proposed method provides improved tracking stability and accuracy when compared to the standard Mean Shift tracking. Furthermore we analyze the proposed method in terms of robustness by assessing its performance in scenarios where occlusions are present. 1